Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Add environment variable
.claude/skills/elie222-environment-variables/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -46% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -17% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -53% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -32% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -18% | 0% |
This is how we add environment variables to the project:
.env.example:bash NEW_VARIABLE=value_example
apps/web/env.ts:typescript // For server-only variables server: { NEW_VARIABLE: z.string(), } // For client-side variables client: { NEXT_PUBLIC_NEW_VARIABLE: z.string(), } experimental__runtimeEnv: { NEXT_PUBLIC_NEW_VARIABLE: process.env.NEXT_PUBLIC_NEW_VARIABLE, }
NEXT_PUBLIC_client and experimental__runtimeEnv sectionsturbo.json under globalDependencies:json { "tasks": { "build": { "env": [ "NEW_VARIABLE" ] } } }
examples:
# Adding a server-side API key # .env.example API_KEY=your_api_key_here
# env.ts server: { API_KEY: z.string(), }
# turbo.json "build": { "env": "API_KEY"] } output: "Server-side environment variable properly added"
# Adding a client-side feature flag # .env.example NEXT_PUBLIC_FEATURE_ENABLED=false
# env.ts client: { NEXT_PUBLIC_FEATURE_ENABLED: z.coerce.boolean().default(false), }, experimental__runtimeEnv: { NEXT_PUBLIC_FEATURE_ENABLED: process.env.NEXT_PUBLIC_FEATURE_ENABLED, }
# turbo.json "build": { "env": "NEXT_PUBLIC_FEATURE_ENABLED"] } output: "Client-side environment variable properly added"
references:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 13,694 | 8,975 | -34% | 1 | 1 | 0% | 2,556 | 1,374 | -46% | 0 | 0 | — |
case-02 | fail→pass | 9,303 | 4,567 | -51% | 1 | 1 | 0% | 1,754 | 1,459 | -17% | 0 | 0 | — |
case-03 | fail→pass | 14,102 | 3,695 | -74% | 1 | 1 | 0% | 2,494 | 1,170 | -53% | 0 | 0 | — |
case-04 | fail→pass | 12,814 | 5,325 | -58% | 1 | 1 | 0% | 2,406 | 1,630 | -32% | 0 | 0 | — |
case-05 | fail→pass | 11,446 | 5,520 | -52% | 1 | 1 | 0% | 2,076 | 1,694 | -18% | 0 | 0 | — |
case-06 | fail→pass | 12,199 | 7,202 | -41% | 1 | 1 | 0% | 2,200 | 1,990 | -10% | 0 | 0 | — |
case-07 | fail→pass | 7,769 | 4,747 | -39% | 1 | 1 | 0% | 1,532 | 1,458 | -5% | 0 | 0 | — |
case-08 | pass→pass | 7,141 | 3,338 | -53% | 1 | 1 | 0% | 1,278 | 1,173 | -8% | 0 | 0 | — |
case-09 | fail→pass | 5,675 | 4,190 | -26% | 1 | 1 | 0% | 1,140 | 1,285 | +13% | 0 | 0 | — |
case-10 | pass→pass | 13,359 | 4,198 | -69% | 1 | 1 | 0% | 2,449 | 1,324 | -46% | 0 | 0 | — |
case-11 | pass→pass | 2,066 | 1,803 | -13% | 1 | 1 | 0% | 338 | 788 | +133% | 0 | 0 | — |
case-12 | fail→pass | 5,986 | 3,840 | -36% | 1 | 1 | 0% | 1,145 | 1,266 | +11% | 0 | 0 | — |
case-13 | fail→pass | 6,751 | 3,223 | -52% | 1 | 1 | 0% | 1,230 | 1,074 | -13% | 0 | 0 | — |
case-14 | pass→pass | 5,688 | 2,278 | -60% | 1 | 1 | 0% | 933 | 894 | -4% | 0 | 0 | — |
case-15 | fail→pass | 7,031 | 4,349 | -38% | 1 | 1 | 0% | 1,296 | 1,307 | +1% | 0 | 0 | — |
case-16 | pass→pass | 4,540 | 2,359 | -48% | 1 | 1 | 0% | 800 | 920 | +15% | 0 | 0 | — |
case-17 | pass→pass | 13,018 | 8,465 | -35% | 1 | 1 | 0% | 2,201 | 1,727 | -22% | 0 | 0 | — |
case-18 | pass→pass | 12,647 | 5,997 | -53% | 1 | 1 | 0% | 1,768 | 1,515 | -14% | 0 | 0 | — |
case-19 | pass→pass | 6,379 | 2,898 | -55% | 1 | 1 | 0% | 1,241 | 1,107 | -11% | 0 | 0 | — |
case-20 | pass→pass | 9,259 | 5,154 | -44% | 1 | 1 | 0% | 1,864 | 1,421 | -24% | 0 | 0 | — |
case-21 | pass→fail | 10,609 | 4,857 | -54% | 1 | 1 | 0% | 1,510 | 1,399 | -7% | 0 | 0 | — |
case-22 | pass→fail | 10,391 | 4,401 | -58% | 1 | 1 | 0% | 2,003 | 1,486 | -26% | 0 | 0 | — |
case-23 | pass→fail | 11,090 | 9,083 | -18% | 1 | 1 | 0% | 2,037 | 1,950 | -4% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 23 cases were attempted. The headline lift of +35 percentage points is the difference between those two pass rates over the 23 comparable cases. 3 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.